The Sideline Captions Constitution

Our position on AI in photography

Preamble

Society is in the middle of an AI reckoning, and photography is no exception. Wire services, photo contests, and press associations are drawing hard lines, and many of them banning AI outright at every step. We understand the fear behind those lines, and we share it. A faked moment or an unverified claim attached to a real photograph is a serious breach of trust. And trust is the only currency a photograph has.

Sideline Captions is built by photographers, for photographers. We revere this craft. The part of photography that matters most is profoundly human: the eye that reads a play before it unfolds, the timing, the instinct, the years of expertise compressed into a single frame – a single moment in time. No software will ever replace it. That is exactly why the flood of AI-generated "art" deeply worries us. When machine-made images are passed off as the real thing, they cheapen the creativity, expertise, and hard-won judgment that real photographers bring, and they blur the line between a real moment that was witnessed and words typed into a text box – a "moment" that was merely conjured. The human piece of this equation is not just decoration. It is precisely what makes an image mean something, and we will never build anything that erodes that.

What we believe about Artificial Intelligence

Not all AI is the same, and treating it as one thing does the craft of photography a disservice. "AI in photography" has come to describe two opposite things; unfortunately, they are often treated as one.

The first kind of AI invents. It generates pixels that were never photographed. It alters what happened, fills in what wasn't there, and can fabricate a moment whole cloth. We believe that this kind of AI has no place in journalism, and we will never build it.

The second kind of AI assists. It reads what is already in the frame and helps with the tedious work that surrounds a photograph – never the photograph itself. Sideline Captions is this second kind.

There is a real and permanent difference between a tool that generates something that never existed and a tool that helps you do faster the work you were always going to do by hand. The first erodes trust. The second protects the thing that earns it — your time, your attention, and your eye for the moment.

While we do have concerns about the rise of AI and what it means for the photographic and journalistic community, we do see a place for AI in society. Used responsibly, AI can be a powerful and incredibly useful tool to augment the important work photographers do day in and day out, and we want to embrace it as such.

Yes, Sideline Captions uses AI, but not in an effort to replace the photographer or to alter the reality of what they have captured. Our use of AI is designed to augment and assist the artist, to free up their time and reduce the mental strain that accompanies captioning photos accurately and efficiently.

This Constitution is our line in the sand. It is available for anyone with an interest in photography and what AI means for the photographic industry: editors deciding whether to trust a tool like Sideline, photographers deciding whether to use it as part of their workflow, and the public the photograph ultimately serves.

So without further ado, this is what we stand for and believe:


Article I — We never let AI alter the image.

Sideline Captions will never use AI to change your photographs. We do not add any objects, nor do we remove any. We do not use generative fill, reframing, or any other type of "AI enhancement". In short, we do nothing that invents or erases content in the frame. Everything that our AI writes is text: the caption and the metadata that travels with the file.

Sideline Live includes basic toning tools — exposure, contrast, temperature, and cropping — but these are the ordinary, hands-on adjustments photographers have always made in the darkroom and with software. The photographer makes every one of these edits yourself; no AI is involved. They refine how a true image is presented; they do not alter what happened in front of the lens. What the camera recorded is locked at the moment of capture, and we never let a machine touch it.

Article II — We describe what is in the frame. We do not invent what isn't.

Sideline Captions reads the visible facts of a photograph, such as a jersey number or the action in front of the lens, and drafts language to describe them. It does not assert a final score it cannot see, an outcome it cannot know, or a person who is not in the frame. It aims to describe, not editorialize. AI sometimes hallucinates, but it is our stated goal to use only the photo and the context provided to the AI to keep hallucinations and fabrications to an absolute minimum.

Article III — The software drafts; the human verifies.

Sideline Captions is a tool. It is fast, and it is fallible. It can misread a number or misidentify a player, just as humans do from time to time. It can describe an action incorrectly. We do not hide this. We design around it. Every caption Sideline Captions produces is a draft, and we treat it as unverified source material until a person confirms it. The photographer's judgment is not a formality at the end of our process: it is the point of our process.

Article IV — A person should always sign the caption.

Sideline Captions never transmits your photos of its own accord, and it never publishes a caption automatically without giving a human a chance to review it first. No image leaves your hands because a machine decided it was ready. In keeping with the editorial and journalistic standards of major publications and wire services, a person should review every caption for accuracy before it is sent. The byline is yours; so is the final responsibility.

Article V — The photographer, not the tool, is the author and the artist.

Sideline is built by photographers. We believe photography is an art. The irreplaceable part of photography — the eye, the timing, the instinct for the moment — belongs to the person behind the camera, not to any software. Photographs are art; captions are not. Captions are grunt work. Nobody got into this craft to spend the night typing names next to numbers. We built Sideline to take that work off photographers' plate and give them hours back, so they can spend less time at their laptop and more time making the art only they can make. We exist to protect the artist's time, never to replace the artist.

Article VI — We tell the truth.

To the best of our ability, we will be honest about what the tool does and where it can be wrong. We will never dress assistance up as authority. Sideline Captions reads the jersey numbers in your frame, matches them against your rosters, and uses AI to connect each number to the correct player and to draft a description of the action. Every part of that is a starting point for the photographer's review, never a verdict. We would rather tell you plainly where a tool can fail than let you trust it more than you should.


What Sideline will never do

The following commitments do not and will not change — not for any customer, at any price:

  • We will never use AI to generate, alter, or manipulate any part of the image itself.
  • We will never add or remove people or objects from a photograph.
  • We will never auto-publish or auto-transmit a caption without giving a human a chance to review it.
  • To the extent that it is avoidable, we will never intentionally invent a fact or name a person that isn't visible in the frame.

Where we stand alongside the industry

The AI bans coming out of the major wire services and photo organizations are aimed squarely at the first kind of AI mentioned earlier in this Constitution – the kind that creates or alters images or fabricates reality. If you read closely, those same standards describe almost exactly how Sideline Captions already works. A few examples:

  • The Associated Press does not allow generative AI to add or subtract elements from a photo, and treats any AI output as unvetted source material that a journalist must edit and stand behind. Its News Values are explicit: AP will not alter photo or image content.
  • Reuters, in its published AI principles, commits to maintaining meaningful human involvement throughout its use of AI.
  • Getty Images prohibits AI-based post-production on editorial images, but states plainly that AI may be used to help create caption text, provided there is human review and verification and the caption meets its accuracy standards.
  • World Press Photo is blunt that an AI-generated image is not a photograph. A photograph is a record of light on a sensor, a real physical moment – and World Press Photo bars adding, removing, or altering content within the frame, while explicitly permitting basic toning and cropping.
  • The NPPA holds that the content of a photograph is locked the instant it is recorded, and its 2026 standards expressly prohibit using generative AI to add, remove, expand, or alter an image.
  • AFP, whose global news director helped author the Paris Charter on AI and Journalism, holds that AI may assist the work but that human journalists remain accountable for what is published, and that images must never be manipulated to give a false picture of events.

Sideline stands with these respected global journalistic bodies, and we strive to meet these standards. In the places that matter most — never using AI to alter the image, never inventing a fact, always keeping a human accountable for the final caption — we hold ourselves to them exactly.

Let's talk

AI represents a sea change in our society. It's a big issue, a thorny topic – one that merits deep discussion and passionate debate. As such, we welcome any thoughts or feedback from our users on the use and appropriate role of AI in photography. If you have any questions about our Constitution, our philosophy on AI, or our approach to using AI as part of the photographic process, we welcome it. Please email us at hello@sidelinecaptions.com. We would love to continue the conversation with you.